US6504201B1 - Memory cell having a vertical transistor with buried source/drain and dual gates - Google Patents

Abstract

An integrated circuit and fabrication method includes a memory cell for a dynamic random access memory (DRAM). Vertically oriented access transistors are formed on semiconductor pillars on buried bit lines. Buried first and second gates are provided for each access transistor on opposing sides of the pillars. Buried word lines extend in trenches orthogonal to the bit lines. The buried word lines interconnect ones of the first and second gates. In one embodiment, unitary gates are interposed and shared between adjacent pillars for gating the transistors therein. In another embodiment, separate split gates are interposed between and provided to the adjacent pillars for separately gating the transistors therein. In one embodiment, the memory cell has a surface area that is approximately 4 F², where F is a minimum feature size. Bulk-semiconductor and semiconductor-on-insulator (SOI) embodiments are provided.

Description

CROSS REFERENCE TO RELATED APPLICATIONS

This application is a continuation of U.S. patent application Ser. No. 08/889,462, filed on Jul. 8, 1997, now U.S. Pat. No. 6,150,687, the specification of which is incorporated herein by reference.

FIELD OF THE INVENTION

The present invention relates generally to integrated circuits, and particularly, but not by way of limitation, to a dynamic random access memory (DRAM) having a memory cell with a vertical access transistor with buried dual gates, and having buried bit and word lines.

BACKGROUND OF THE INVENTION

Semiconductor memories, such as dynamic random access memories (DRAMs), are widely used in computer systems for storing data. A DRAM memory cell typically includes an access field-effect transistor (FET) and a storage capacitor. The access FET allows the transfer of data charges to and from the storage capacitor during reading and writing operations. The data charges on the storage capacitor are periodically refreshed during a refresh operation.

Memory density is typically limited by a minimum lithographic feature size (F) that is imposed by lithographic processes used during fabrication. For example, the present generation of high density dynamic random access memories (DRAMs), which are capable of storing 256 Megabits of data, require an area of 8 F²per bit of data. There is a need in the art to provide even higher density memories in order to further increase data storage capacity and reduce manufacturing costs. Increasing the data storage capacity of semiconductor memories requires a reduction in the size of the access FET and storage capacitor of each memory cell. However, other factors, such as subthreshold leakage currents and alpha-particle induced soft errors, require that larger storage capacitors be used. Thus, there is a need in the art to increase memory density while allowing the use of storage capacitors that provide sufficient immunity to leakage currents and soft errors. There is also a need in the broader integrated circuit art for dense structures and fabrication techniques.

SUMMARY OF THE INVENTION

The present invention provides an integrated circuit including a pillar of semiconductor material that extends outwardly from a working surface of a substrate. The pillar has a number of sides. A transistor is formed having a body region and first and second source/drain regions within the pillar. The transistor includes first and second gates that are each associated with a side of the pillar.

The invention also provides a memory device including an array of memory cells. Each cell includes a transistor. Each transistor includes a semiconductor pillar forming body and first and second source/drain regions. The transistor also includes first and second gates disposed adjacent to opposing sides of the pillar. The memory device also includes a plurality of substantially parallel first word lines. Each first word line is disposed orthogonally to the bit lines in a trench between columns of the memory cells. Each first word line allows addressing of first gates of the transistors of the memory cells that are adjacent to the trench in which the first word line is disposed. The memory device also includes a plurality of substantially parallel second word lines. The second word lines are interdigitated with the first word lines. Each second word line is disposed orthogonally to the bit lines in a trench between columns of the memory cells. Each second word line allows addressing of second gates of the transistors of the memory cells that are adjacent to the trench in which the second word line is disposed. A plurality of bit lines is provided, proximal to the substrate. The bit lines interconnect ones of the first source/drain regions of ones of the memory cells. In one embodiment, the pillars extend outwardly from an insulating portion of the substrate. In another embodiment, the pillars extend outwardly from a semiconductor portion of the substrate.

The invention also provides a method of fabricating an integrated circuit. According to one embodiment of the method, a substrate is provided, and a plurality of bit lines are formed on the substrate. A plurality of access transistors are formed on each of the bit lines. Each access transistor includes a first source/drain region shared by at least a portion of the bit line. Each access transistor also includes a body region and second source/drain region formed vertically on the first source/drain region. A plurality of isolation trenches are formed in the substrate, orthogonal to the bit lines. Each trench is located between access transistors on the orthogonal bit lines. A first word line is formed in a first one of the trenches. The first word line controls conduction between first and second source/drain regions of access transistors that are adjacent to a first side of the first trench. A second word line is formed in a second one of the trenches. The second word line controls conduction between first and second source/drain regions of access transistors that are adjacent to a first side of the second trench.

In one embodiment, the first word line also controls conduction between first and second source/drain regions of access transistors that are adjacent to a second side of the first trench. In another embodiment, the second word line also controls conduction between first and second source/drain regions of access transistors that are adjacent to a second side of the second trench.

In a further embodiment, another first word line is formed in the first trench, for controlling conduction between first and second source/drain regions of access transistors that are adjacent to a second side of the first trench. In yet a further embodiment, a second word line is formed in the second trench, for controlling conduction between first and second source/drain regions of access transistors that are adjacent to a second side of the second trench.

Thus, the invention provides high density integrated circuit structures and fabrication methods, such as for DRAM memory cell arrays and other semiconductor devices. Each memory cell can be fabricated in a surface area that is approximately 4 F², where F is a minimum lithographic feature size. In one embodiment, a common first word line is shared by all of the access FETs that are located along both sides of the trench in which the first word line is located. In another embodiment, a common second word line is shared by all of the access FETs that are located along both sides of the trench in which the second word line is located. In further embodiments, split word lines are provided in either or both trenches. The split word lines provide separate addressing of gate regions of access FETs on opposite sides of the trench. Each of the unitary and split word line embodiments can be fabricated on a bulk semiconductor substrate, or on a semiconductor-on-insulator (SOI) substrate that results from using an SOI starting material, or by forming SOI regions during fabrication. The SOI embodiments provide greater immunity to alpha-particle induced soft errors, allowing the use of smaller storage capacitors.

BRIEF DESCRIPTION OF THE DRAWINGS

In the drawings, like numerals describe substantially similar components throughout the several views.

FIG. 1 is a schematic representation illustrating generally one embodiment of a semiconductor memory according to the invention.

FIG. 2 is a perspective view illustrating generally one embodiment of a portion of a memory according to the present invention.

FIG. 3 is a plan view illustrating generally memory cells according to one embodiment of the invention as viewed from above the structures formed on the substrate.

FIG. 4 is a cross-sectional view taken along thecut line4—4 of FIG.3.

FIGS. 5A-K describe generally various processing techniques of one embodiment of a method of fabricating memory cells according to the invention.

FIG. 6 is a perspective view illustrating generally another embodiment of a portion of a memory according to the present invention having a semiconductor-on-insulator (SOI) substrate.

FIGS. 7A-C illustrate generally, by way of example, additional steps used to form SOI bars according to one embodiment of the invention.

FIG. 8 is a perspective view illustrating-generally another embodiment of a portion of a memory according to the present invention in which split gates are formed.

FIG. 9 is a plan view illustrating generally memory cells according to a split gate embodiment of the invention as viewed from above the structures formed on the substrate.

FIG. 10 is a cross-sectional view taken along thecut line10—10 of FIG.9.

FIG. 11 is a perspective view illustrating generally another embodiment of a portion of a memory according to the present invention including split gates and a bulk semiconductor substrate.

FIG. 12 is a cross-sectional view taken along thecut line12—12 of FIG.11.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

In the following detailed description of the invention, reference is made to the accompanying drawings which form a part hereof, and in which is shown, by way of illustration, specific embodiments in which the invention may be practiced. The embodiments are intended to describe aspects of the invention in sufficient detail to enable those skilled in the art to practice the invention. Other embodiments may be utilized and changes may be made without departing from the scope of the present invention. In the following description, the terms wafer and substrate are interchangeably used to refer generally to any structure on which integrated circuits are formed, and also to such structures during various stages of integrated circuit fabrication. Both terms include doped and undoped semiconductors, epitaxial layers of a semiconductor on a supporting semiconductor or insulating material, combinations of such layers, as well as other such structures that are known in the art. The following detailed description is not to be taken in a limiting sense, and the scope of the present invention is defined only by the appended claims.

FIG. 1 is a schematic representation illustrating generally one embodiment of anintegrated circuit100, such as a semiconductor memory device, incorporating an array of memory cells provided by the invention. In FIG. 1,circuit100 illustrates, by way of example but not by way of limitation, a dynamic random access memory (DRAM), but the invention also comprises other integrated circuits including any other semiconductor memory devices. In this exemplary embodiment,circuit100 includes memory cell arrays110, such as110A and110B. Each array110 includes M rows and N columns ofmemory cells112.

In the exemplary embodiment of FIG. 1, each memory cell includes a transfer device, such as n-channel cell access field-effect transistor (FET)130 or any other transistor or switching device having more than one control terminal inputs. More particularly,access FET130 includes first and second gate terminals for controlling conduction between its first and second source/drain terminals.

Access FET130 is coupled at a second source/drain terminal to a storage node of astorage capacitor132. The other terminal ofstorage capacitor132 is coupled to a reference voltage such as a ground voltage VSS. Each of the M rows includes one of word lines WL0, WL1 . . . WLm−1, WLm coupled to the first gate terminals ofaccess FETs130 or to one of the control terminals of an equivalent switching device. Each of the M rows also includes one of word lines R1, R2, . . . , Rm−1, Rm coupled to the second gate terminals ofaccess FETs130 inmemory cells112. Thus, the term “word line” includes any interconnection line between gate terminals ofaccess FETs130, or the control terminals of equivalent switching devices. Each of the N columns includes one of bit lines BL0, BL1 . . . BLn−1, BLn.

Bit lines BL0-BLn are used to write to and read data frommemory cells112. Word lines WL0-WLm and R1-Rm are used to activateaccess FETs130 to access a particular row ofmemory cells112 that is to be written or read. Addressing circuitry is also included. For example,address buffer114controls column decoders118, which also include sense amplifiers and input/output circuitry that is coupled to bit lines BL0-BLn.Address buffer114 also controlsrow decoders116.Row decoders116 andcolumn decoders118 selectablyaccess memory cells112 in response to address signals that are provided onaddress lines120 during read and write operations. The address signals are typically provided by an external controller such as a microprocessor or other memory controller. Each ofmemory cells112 has a substantially identical structure, and accordingly, only onememory cell112 structure is described herein.

In one example mode of operation,circuit100 receives an address of aparticular memory cell112 ataddress buffer114.Address buffer114 identifies one of the word lines WL0-WLm of theparticular memory cell112 to rowdecoder116.Row decoder116 selectively activates the particular word line WL0-WLm to activateaccess FETs130 of eachmemory cell112 that is connected to the selected word line WL0-WLm.Column decoder118 selects the one of bit lines BL0-BLn of the particularly addressedmemory cell112. For a write operation, data received by input/output circuitry is coupled to the one of bit lines BL0-BLn and through theaccess FET130 to charge or discharge thestorage capacitor132 of the selectedmemory cell112 to represent binary data. For a read operation, data stored in the selectedmemory cell112, as represented by the charge on itsstorage capacitor132, is coupled to the one of bit lines BL0-BLn, amplified, and a corresponding voltage level is provided to the input/output circuits.

According to one aspect of the invention, each of the first and second gates ofaccess FET130 is capable of controlling the conduction between its first and second source/drain terminals, as described below. In this embodiment, parallel switching functionality can be effected between the first and second source/drain terminals ofaccess FET130 by independently operating the particular ones of word lines WL0-WLm and corresponding ones of word lines R0-Rm. For example, by independently activating word line WL0 and word line R0, both of which are coupled to the same row ofmemory cells112, independently controlled inversion channels can be formed in eachcorresponding access FET130 by respective first and second gates for allowing conduction between the first and second source/drain regions.

According to another aspect of the invention, each of the first and second gates ofaccess FET130 is capable of controlling the conduction between its first and second source/drain terminals, but the first and second gates ofparticular access FETs130 are synchronously activated, rather than independently operated. For example, by synchronously activating word line WL0 and word line R0, both of which are coupled to the same row ofmemory cells112, synchronously activated inversion channels can be formed in eachcorresponding access FET130 by respective first and second gates for allowing conduction between the first and second source/drain regions.

In this embodiment, synchronous activation and deactivation of the first and second gates allows better control over the potential distributions in theaccess FET130 when it is in a conductive state. Synchronous activation and deactivation can be used to obtain well-controlled fully depleted operating characteristics ofaccess FET130.

In a further embodiment in which the first and second gates are either synchronously or independently activated, different activation voltages can be applied to the first and second gates of theaccess FET130. For example, different voltages can be provided to synchronously activated word lines WL0 and R0, thereby providing different activation voltages to the first and second gates of theaccess FET130 to obtain particular desired operating characteristics. Similarly, different deactivation voltages can be applied to the first and second gates of theaccess FET130. For example, different deactivation voltages can be provided to synchronously deactivated word lines WL0 and R0 and corresponding first and second gates ofaccess FETs130, in order to obtain particular desired operating characteristics. Similarly, different activation and deactivation voltages can be applied to independently operated word lines such as WL0 and R0.

FIG. 2 is a perspective view illustrating generally one-embodiment of a portion of a memory according to the present invention. FIG. 2 illustrates portions of sixmemory cells112a-f, including portions of vertically orientedaccess FETs130 therein. Conductive segments ofbit lines202 represent particular ones of bit lines BL0-BLn. Conductive segments offirst word line206 represents any one of word lines WL0-WLm, which provide integrally formed first gates foraccess FETs130 between which the particularfirst word line206 is interposed. Conductive segments ofsecond word line208 represents any one of word lines R0-Rm, which provide integrally formed second gates foraccess FETs130 between which the particularsecond word line208 is interposed. Thus, word lines WL0-WLm and R0-RM are alternatingly disposed (interdigitated) within the array110. The detailed description ofmemory cell112 structure refers only tomemory cells112a-f,bit lines202, and respective first and second word lines206 and208 that are associated withmemory cells112a-f. However, the following description similarly applies to allmemory cells112 and similar conductive lines in array110.

In FIG. 2, vertically orientedaccess FETs130 are formed in semiconductor pillars that extend outwardly from anunderlying substrate210. As described below,substrate210 includes bulk semiconductor starting material, semiconductor-on-insulator (SOI) starting material, or SOI material that is formed from a bulk semiconductor starting material during processing.

In one example embodiment, using bulk silicon processing techniques,access FETs130 include an n+silicon layer formed on abulk silicon substrate210 to produce first source/drain regions212 ofaccess FETs130 and integrally formed n++ conductively dopedbit lines202 defining a particular column ofmemory cells112. A p− silicon layer is formed on n+ first source/drain region212 to form thebody region214 ofaccess FET130, in which inversion channels may be capacitively generated at the sidewalls of the semiconductor pillar under the control of the first and second gates. A further n+ silicon layer is formed on p-body region214 to produce second source/drain region216 of access FET130:Storage capacitors132 are formed on the second/source drain regions216.

Word lines WL0-WLm and R0-RM are alternatingly disposed (interdigitated) within the array110. For example,first word line206 is interposed between semiconductor pillars of memory cell pairs112a-band112d-e.Second word line208 is interposed between semiconductor pillars of memory cell pairs112b-cand112e-f. Thus, as seen from FIG. 2,access FETs130 are formed onbit lines202 in semiconductor pillars extending outwardly fromsubstrate210 and includingbody regions214, and first and secondsource drain regions212 and216, respectively. In this embodiment,bit lines202 contactbulk semiconductor substrate210.

Isolation trenches provide isolation betweenaccess FETs130 ofadjacent memory cells112. Columns ofmemory cells112 are separated by atrench220 that is subsequently filled with a suitable insulating material such as silicon dioxide. For example,trench220 provides isolation betweenmemory cells112aand112dand betweenmemory cells112band112e. Rows ofmemory cells112 are alternatingly separated by atrench221 and222, each of which are separated fromsubstrate210 by an underlying insulating layer, described below, and separated from thebody region214 ofaccess FETs130 by a gate oxide, also described below. For example,trench221 provides isolation betweenmemory cells112aand112band betweenmemory cells112dand112e. In addition,trench222 provides isolation betweenmemory cells112band112candmemory cells112eand112f.Trenches221 and222 extend substantially orthogonally to bitlines202.

FIG. 3 is a plan view illustrating generallymemory cells112a-fas viewed from above the structures formed onsubstrate210. FIG. 3 illustrates subsequently formed insulator such asoxide224, formed intrenches220 to provide isolation betweenmemory cells112. In this embodiment,first word line206 is shared between first gates ofaccess FETs130 ofmemory cells112 in adjacent rows, such as betweenmemory cells112a-band112d-e.First word line206 is also shared between first gates ofother access FETs130 that are in the same adjacent rows, but coupled to different bit lines202.First word line206 is located in-trench221 that extends between the semiconductor pillars ofmemory cells112aand112b.First word line206 is separated bygate oxide218 from thevertical sidewalls219 of the semiconductor pillars on each side oftrench221.

Second word line208 is shared between second gates ofaccess FETs130 ofmemory cells112 in adjacent rows, such as betweenmemory cells112b-cand112e-f.Second word line208 is also shared between second gates ofother access FETs130 that are in the same adjacent rows, but coupled to different bit lines202.Second word line208 is located intrench222 that extends between the semiconductor pillars ofmemory cells112band112c.Second word line208 is separated bygate oxide218 from thevertical sidewalls223 of the semiconductor pillars on each side oftrench222.

FIG. 4 is a cross-sectional view taken along thecut line4—4 of FIG.3. In FIG. 4, respective first and second word lines206 and208 are buried below theactive semiconductor surface230 of the semiconductor pillar in thememory cells112.Active semiconductor surface230 represents an upper semiconductor portion of second source/drain region216. First and second word lines206 and208, respectively, are isolated from adjacent semiconductor pillars bygate oxide218. First and second word lines206 and208, respectively provide integrally formed first and second gate portions that are capacitively coupled toadjacent access FET130body regions214, such as for forming inversion channel regions therein. In one embodiment, respective first and second word lines206 and208 are formed of a refractory metal, such as tungsten or titanium. In another embodiment, first and second word lines206 and208 can be formed of n+ doped polysilicon. Similarly, other suitable conductors could also be used for first andsecond words lines206 and208, respectively.

Burying first and second word lines206 and208 belowsemiconductor surface230 provides additional space on the upper portion ofmemory cell112 for formation ofstorage capacitors132. Increasing the area available for formingstorage capacitor132 increases the possible obtainable capacitance value ofstorage capacitor132. In one embodiment,storage capacitor132 is a stacked capacitor that is formed using any of the many capacitor structures and process sequences known in the art. Other techniques could also be used for implementingstorage capacitor132. Contacts to the first and second word lines206 and208, respectively, can be made outside of the memory array110.

As illustrated in the plan view of FIG. 3, respective first and second word lines206 and208 are shared betweenadjacent memory cells112. As a result, only one-half the surface line width of each is allocated to each memory cell. The row pitch of each cell, measured from the centerline offirst word line206 to the centerline ofsecond word line208, can be approximately 2 F, where F is a minimum lithographic feature size. F corresponds to the length and width presented by the surface of a minimum-sized semiconductor pillar in eachmemory cell112. The column pitch of each cell, measured between centerlines ofbit lines202 can be approximately 2 F. Thus, the surface area of eachmemory cell112 can be approximately 4 F².

FIGS. 5A-K describe generally various processing techniques of one embodiment of a method of fabricatingmemory cells112, such as shown in FIGS. 2-4, using bulk silicon processing techniques. In the embodiment of FIG. 5A, a p-bulk silicon substrate210 starting material is used. An n++ and n+ silicon composite first source/drain layer212 is formed onsubstrate210, such as by ionimplantation, epitaxial growth, or a combination of such techniques. The more heavily conductively doped lower portion of the first/source drain layer212 also functions as thebit line202. The thickness of the n++ portion of first source/drain layer212 is that of the desiredbit line202 thickness, which can be approximately between 0.1 to 0.25 μm. The overall thickness of the first source/drain layer212 can be approximately between 0.2 to 0.5 μm. Abody region layer214 of p− silicon is formed, such as by epitaxial growth, to a thickness that can be about 0.4 μm. A second source/drain region layer216 of n+ silicon is formed, such as by ion-implantation intobody region layer214 or by epitaxial growth onbody region layer214, to a thickness that can be approximately between 0.2 and 0.5 μm.

In FIG. 5B, an SiO₂thinpad oxide layer512 is formed on second source/drain region216, such as by chemical vapor deposition (CVD). In one embodiment,pad oxide layer512 can be approximately 10 nm in thickness. A thin silicon nitride (Si₃N₄)layer514 is formed onpad oxide layer512, such as by CVD. In one embodiment,nitride layer514 of can be approximately 100 nm in thickness.

In Figure SC, photoresist is applied and selectively exposed to provide a mask for the directional etching oftrenches220, such as by reactive ion etching (RIE). The directional etching results in a plurality of column bars516 containing the stack ofnitride layer514,pad oxide layer512, second source/drain layer216,body region layer214, and first source/drain layer212.Trenches220 are etched to a depth that is sufficient to reach thesurface518 ofsubstrate210, thereby providing separation between conductively doped bit lines202.Bars516 are oriented in the direction of bit lines202. In one embodiment, bars516 have a surface line width of approximately one micron or less. The depth and width of eachtrench220 can be approximately equal to the line width ofbars516.

In FIG. 5D, the photoresist is removed.Isolation material224, such as SiO₂is deposited to fill thetrenches220. The working surface is then planarized, such as by chemical mechanical polishing/planarization (CMP).

FIG. 5E illustrates the view of FIG. 5D after clockwise rotation by ninety degrees. In FIG. 5E, a photoresist material is applied and selectively exposed to provide a mask for the directional etching oftrenches221 and222, such as by reactive ion etching (RIE) of a plurality of row bars532 that are disposed orthogonally to bitlines202. Formingtrenches221 and222 includes etching though stacked layers in the portions ofbars516. Formingtrenches221 and222 also includes etching through theisolation material224 disposed betweenbars516.

More particularly,trenches221 and222 are etched throughnitride layer514,pad oxide layer512, second source/drain layer216,body region layer214, and partially into first source/drain layer212.Trenches221 and222 are etched intobars516 to a depth of about 100 nm into first source/drain layer212, leaving intact anunderlying bit line202 portion of the first source/drain layer212.Trenches221 and222 are also etched into theisolation material224 betweenbars516. In one embodiment, after etchingnitride layer514 ofbars516, a nonselective dry etch is used to remove theisolation material224 betweenbars516 and also thepad oxide layer512, second source/drain layer216,body region layer214, and a portion of first source/drain layer212 ofbars516. The directional etching oftrenches221 and222 results in the formation of a plurality of row bars532 that are orthogonal to column bars516.

FIG. 5F is a plan view illustrating generally the arrangement ofparallel bars516, andtrenches220 interposed therebetween.Bars532 are arranged orthogonally tobars516.Trenches221 and222 are interposed between ones ofbars532. The resulting semiconductor pillars in the intersecting portions ofbars532 and516 provide first and secondsource drain regions212 and216, respectively, andbody region214 foraccess FETs130 ofmemory cells112.

In FIG. 5G, which is oriented similarly to FIG. 5E, a conformalsilicon nitride layer540 is formed, such as by CVD.Nitride layer540 is directionally etched, such as by RIE, to leave resulting portions ofnitride layer540 only onsidewalls219 of thebars532 intrenches221 and222. In one embodiment, the thickness ofnitride layer540 is about 20 mn. Anoxide layer542 is formed, such as by thermal growth, at the base portions oftrenches221 and222.Oxide layer542 insulates theunderlying bit lines202 from structures subsequently formed intrenches221 and222. After formingoxide layer542, remaining portions ofnitride layer540 are removed.

In FIG. 5H, agate oxide218 is formed on the exposed sidewalls219 portions intrenches221 and222 of second source/drain region216,body region214, and first source/drain region212. In one embodiment,gate oxide218 is a high-quality thin oxide layer that is thermally grown on the exposed sidewalls219 oftrenches221 and222.

In FIG. 5I, aconductive layer544 is formed over the working surface of the wafer, including fillingtrenches221 and222 in which respective first and second word lines206 and208 will be formed. In one embodiment,layer544 is formed by CVD of a refractory metal, such as tungsten. In another embodiment,layer544 is formed by CVD of n+ polysilicon.

In FIG. 5J, CMP or other suitable planarization process is used to remove portions oflayer544 above the interface betweenpad oxide512 and second source/drain layer216.Pad oxide512 andnitride layer514 are also removed during this planarization step. As a result of the planarization step, first and second word lines206 and208, respectively, are formed inrespective trenches221 and222.

FIG. 5K illustrates one embodiment in which, an insulatinglayer546, such as SiO₂, is formed on the working surface of the wafer, such as by CVD. The structure thus formed is then processed to fabricate astorage capacitor132 on the working surface of the wafer, using known techniques, followed by conventional back end of line (BEOL) procedures.

FIG. 6 is a perspective view illustrating generally another embodiment of a portion of a memory according to the present invention, similar to that described with respect to FIG.2. FIG. 6, however, illustrates an embodiment of the present invention having a semiconductor-on-insulator (SOI)substrate210 rather than abulk semiconductor substrate210. In one embodiment,SOI substrate210 of FIG. 6 is obtained using an SOI starting material. In another embodiment, described below, a bulk semiconductor starting material is used, and an bars of SOI are formed during fabrication ofcircuit100, such that the semiconductor pillars, in which portions ofaccess FETs130 are formed, extend outwardly from an insulatingportion602 ofsubstrate210.

One such method of forming bars of SOI is described in the Noble U.S. patent application Ser. No. 08/745,708 which is assigned to the assignee of the present application and which is herein incorporated by reference. Another such method of forming regions of SOI is described in the Forbes U.S. patent application Ser. No. 08/706,230, which is assigned to the assignee of the present application and which is herein incorporated by reference.

FIGS. 7A-C illustrate generally, by way of example, additional steps used to form SOI bars during the fabrication steps described above with respect to FIGS. 5A-K, such that the semiconductor pillars, in which portions ofaccess FETs130 are formed, extend outwardly from an insulatingportion602 ofsubstrate210, resulting in the structure illustrated in FIG.6.

In FIG. 7A, the processing steps described above with respect to FIGS. 5A-C are carried out, formingtrenches220 that are etched to a depth that is below theoriginal surface518 ofsubstrate210, such as by approximately greater than or equal to 0.6 μm. Anitride layer704 is formed, such as by CVD.Nitride layer704 is directionally etched, such as by RIE, to removenitride layer704 from the base regions oftrenches220. Portions ofnitride layer704 remain on the sidewall oftrenches220 to protect adjacent layers during subsequent etching and oxidation.

In FIG. 7B, an isotropic chemical etch of silicon is used to partially undercut bars516. For example, hydrofluoric acid (HF) or a commercial etchant sold under the trade name CP4 (a mixture of approximately 1 part (46% HF):1 part (CH₃COOH):3 parts (HNO₃)) can be used for the isotropic etchant. In one embodiment, the partial undercutting ofbars516 by isotropic etching is timed to remove a volume of silicon that is sufficient to compensate for a subsequently formed volume of oxide, described below. In general, the subsequent oxidation step produces a volume of oxide that is approximately twice that of the silicon consumed during oxidation.

In FIG. 7C,substrate210 is oxidized using a standard semiconductor processing furnace at a temperature of approximately 900 to 1,100 degrees Celsius. A wet oxidizing ambient is used in the furnace chamber to oxidize the exposed silicon regions in the lower portion oftrenches220.Substrate210 is oxidized for a time period that is sufficient to formoxide insulating portion602 that fully undercutsbars516. Insulatingportion602 underlies bothbars516 andtrenches220, and isolates thebit lines202 andaccess FETs130 formed onbit lines202 from an underlying semiconductor portion ofsubstrate210.Nitride layer704 is removed, and processing then continues as described above with respect to FIGS. 5D-K, resulting in the structure of FIG.6.

In one embodiment, bars516 are sufficiently narrow such that the oxidation step that undercutsbars516 produces sufficient oxide to filltrenches220, resulting in a generally planar structure. This avoids the need for a separate step of depositing anoxide insulation material224 described with respect to FIG.5D. The oxidation time period depends on the width ofbars516 and the effective width ofbars516 after the undercut etch step.Narrower bars516 require shorter oxidation times. For example, for sub-0.25 micron technology, oxidation time is approximately 1 hour. In another embodiment, the etch step fully undercutsbars516 before oxidation. This further reduces oxidation time.

FIGS. 8-10 illustrate generally another embodiment of a portion of a memory according to the present invention, similar to that described with respect to FIG.6. In the embodiment of FIGS. 8-10, however, first and second word lines206 and208, respectively, are each split into separate conductors.First word line206 is split into independently operablefirst word lines206aand206b, each disposed intrench221 and electrically isolated from each other.Second word line208 is split into independently operable second word lines208aand208b, each disposed intrench222 and electrically isolated from each other, such as by SiO₂. Thus, gate regions need not be shared betweenaccess FETs130 inadjacent memory cells112 on opposing sides oftrenches221 and222. First and second word lines206 and208 can be formed of a refractory metal or n+ polysilicon or other suitable conductor, as described above.

In FIGS. 8-10, for example,first word line206aextends intrench221 adjacent to thevertical sidewalls219 of the semiconductor pillars of in-line memory cells112aand112d, separated therefrom bygate oxide218.First word line206bextends intrench221 adjacent to thevertical sidewalls219 of the semiconductor pillars of in-line memory cells112band112e, separated therefrom bygate oxide218.Second word line208aextends intrench222 adjacent to thevertical sidewalls219 of the semiconductor pillars of in-line memory cells112band112e, separated therefrom bygate oxide218.Second word line208bextends intrench222 adjacent to thevertical sidewalls219 of the semiconductor pillars of in-line memory cells112cand112f.

Operation of theaccess FET130 ofmemory cell112b, for example, includes operation of thefirst word line206bandsecond word line208a, as described above. A positive potential is applied to either or both offirst word line206bandsecond word line208a, as described above to turn on theaccess FET130 ofmemory cell112b. However, sincefirst word line206bis not shared by theaccess FET130 ofmemory cell112a, subthreshold leakage is not induced in theaccess FET130 ofmemory cell112aduring activation offirst word line206bto operate theaccess FET130 ofmemory cell112b. Similarly, sincesecond word line208ais not shared by theaccess FET130 ofmemory cell112c, subthreshold leakage in theaccess FET130 ofmemory cell112cis not induced during activation ofsecond word line208ato operate theaccess FET130 ofmemory cell112b.

The use of splitfirst word lines206a-band splitsecond word lines208a-bavoids the problem of sub-threshold conduction inaccess FETs130 in one row while thememory cells112 in the adjacent row are being addressed. Eachmemory cell112 is capable of being uniquely addressed by a combination offirst word line206 andsecond word line208 voltages. These voltages need not appear on thefirst word line206 andsecond word line208 of adjacent rows ofmemory cells112.

The structures of FIGS. 8-10 can be fabricated by process steps similar to those described with respect to FIGS. 5A-C forming bars516 separated bytrenches220. This is followed by the process steps described with respect to FIGS. 7A-C follow, isolatingbars516 from an underlying semiconductor portion ofsubstrate210 by insulatingportion602. This is followed by the process steps described with respect to FIGS. 5D-J forming a planar structure including a unitary conductorfirst word line206 infirst trench221 and a unitary conductorsecond word line208 insecond trench222.

Unitary conductorfirst word line206 thus formed is split to formseparate conductors206a-b. A refractory metal, n+ polysilicon, or other conductor is deposited as a conformal film that can have a thickness of less than or equal to approximately F/3, where F is the minimum feature size. The conformal film is then directionally etched, thereby leaving resulting splitconductor word lines206a-badjacent to thevertical sidewall219, separated therefrom bygate oxide218.Second word line208 can be similarly split intoseparate conductors208a-bduring the same deposition and directional etch steps. Splitting unitary conductor first and second word lines206 and208, respectively, provides the resulting structures illustrated in FIGS. 8-10, but is not required to practice the invention. These steps can be omitted, such as to obtain the structures illustrated in FIGS. 2-4.

FIGS. 11-12 illustrate generally another embodiment of a portion of a memory according to the present invention, similar to that described with respect to FIGS. 8-10, but fabricating using abulk silicon substrate210, as described with respect to FIGS. 2-4. The structures of FIGS. 11-12 are fabricated by process steps similar to those described with respect to FIGS. 5A-C, formingbars516 separated bytrenches220. The process steps described with respect to FIGS. 7A-C follow are omitted from this embodiment. Instead, this is followed by the process steps described with respect to FIGS. 5D-J forming a planar structure including a unitary conductorfirst word line206 infirst trench221 and a unitary conductorsecond word line208 insecond trench222. The unitary conductor first and second word lines206 and208, respectively, are then split as described above with respect to FIGS. 8-10.

The above structures and fabrication methods have been described, by way of example, and not by way of limitation, with respect to memory integrated circuits such as dynamic random access memories (DRAMs). However, the scope of the invention includes any other integrated circuit applications in which the above structures and fabrication methods are used.

Thus, it has been shown that the invention provides integrated circuit structures and fabrication methods, such as for DRAM memory cell arrays and other semiconductor devices. Each memory cell includes a vertical access FET, formed on a bit line, and first and second gates integrally formed with respective first and second word lines that are buried in trenches that extend along opposite sides of the memory cell. Each memory cell can be fabricated in a surface area that is approximately 4F², where F is a minimum lithographic feature size. In one embodiment, a common first word line is shared by all of the access FETs that are located along both sides of the trench in which the first word line is located. Also, a common second word line is shared by all of the access FETs that are located along both sides of the trench in which the second word line is located. In another embodiment, split word lines are provided in each trench, and the two split word lines provide separate addressing of gate regions of access FETs on opposite sides of the trench. Each of the unitary and split word line embodiments can be fabricated on a bulk semiconductor substrate, or on a semiconductor-on-insulator (SOI) substrate that results from using an SOI starting material, or by forming SOI regions during fabrication.

Claims (28)

What is claimed is:

1. A memory array, comprising:

a substrate;

a plurality of semiconductor pillars that extend outwardly from a working surface of the substrate, each pillar having a plurality of sides including a first side opposite a second side, each pillar having a first source/drain region, a body formed vertically on the first source/drain region, and a second source/drain region formed vertically on the body, the first and second source/drain regions and the body of each pillar forming portions of an access transistor for a memory cell, the pillars being arranged in an array including a plurality of columns separated by first trenches and a plurality of rows separated by second trenches extending orthogonally to the first trenches, the first and second sides of the pillars in each row of pillars facing adjacent second trenches;

a plurality of bit lines, each bit line electrically interconnecting one of the first and the second source/drain regions of each of the pillars in one of the columns of the array; and

a pair of electrically-isolated word lines on opposite sides of each second trench, a first word line of each pair disposed adjacent to the first sides of the pillars in the row of pillars on a first side of the respective second trench to form a first gate for those pillars, a second word line of each pair disposed adjacent to the second sides of the pillars in the row of pillars on a second side of the respective second trench to form a second gate for those pillars, whereby each pillar has first and second gates associated with the first and second sides of that pillar, respectively.

2. The memory array ofclaim 1, wherein each bit line is disposed between the bodies of the pillars in the respective column of the array and the substrate.

3. The memory array ofclaim 1, wherein the first source/drain region of each pillar includes a less heavily doped portion and a more heavily doped bit line portion.

4. The memory array ofclaim 1, wherein each pillar extends outwardly from an insulating portion of the substrate.

5. The memory array ofclaim 1, wherein each pillar extends outwardly from a semiconductor portion of the substrate.

6. The memory array ofclaim 1, further comprising a storage capacitor coupled to one of the first and second source/drain regions of each pillar.

7. The memory array ofclaim 1, further comprising a first insulating layer disposed between the first gate and the first side of each pillar, and a second insulating layer disposed between the second gate and the second side of each pillar.

8. A memory device, comprising:

an array of memory cells having a plurality of columns and a plurality of rows, each cell comprising an access transistor including a semiconductor pillar with body and first and second source/drain regions, and first and second gates disposed adjacent to opposing sides of the pillar;

a plurality of bit lines, each bit line electrically interconnecting one of the first and the second source/drain regions of each of the pillars in one of the columns of the array;

a plurality of first word line pairs, the first word line pairs disposed in first trenches extending orthogonal to the bit lines between adjacent rows of cells, each first word line in the first word line pair electrically isolated from the other first word line in the first word line pair; and

a plurality of second word line pairs, interdigitated with the first word line pairs, the second word line pairs disposed in second trenches extending orthogonal to the bit lines between adjacent rows of cells, each second word line in the second word line pair electrically isolated from the other second word line in the second word line pair, wherein the first gate of each access transistor is addressed by one of the first word lines disposed in the first trench that is adjacent to that access transistor, and the second gate of each access transistor is addressed by one of the second word lines disposed in the second trench adjacent to that access transistor.

9. The memory device ofclaim 8, wherein each bit line is disposed between the bodies of the pillars in the respective column of the array and a semiconductor substrate.

10. The memory device ofclaim 8, wherein the first source/drain region of each pillar includes a less heavily doped portion and a more heavily doped bit line portion.

11. The memory device ofclaim 8, further comprising a substrate, wherein each pillar extends outwardly from an insulating portion of the substrate.

12. The memory device ofclaim 8, further comprising a substrate, wherein each pillar extends outwardly from a semiconductor portion of the substrate.

13. The memory device ofclaim 8, wherein each cell further comprises a storage capacitor coupled to one of the first and second source/drain regions of the respective pillar.

14. The memory device ofclaim 8, wherein adjacent columns of cells are separated by third trenches that extend orthogonally to the first and second trenches.

15. An integrated circuit, comprising:

a substrate;

a pillar of semiconductor material that extends outwardly from a surface of the substrate, and that has an upper surface and a plurality of sides including opposing first and second sides;

an access transistor having a vertically-stacked body and first and second source/drain regions formed within the pillar, and also having a first gate associated with the first side of the pillar and a second gate associated with the second side of the pillar;

a storage capacitor coupled to a first one of the first and second source/drain regions;

a bit line coupled to a second one of the first and second source/drain regions;

a first word line disposed in a first trench extending orthogonal to the bit line on the first side of the pillar, wherein the first word line provides control of the first gate; and

a second word line disposed in a second trench extending orthogonal to the bit line on the second side of the pillar, the second word line being electrically isolated from the first word line, wherein the second word line provides control of the second gate.

16. The integrated circuit ofclaim 15, wherein the first and second word lines are disposed beneath the upper surface of the pillar, and the storage capacitor is disposed above the upper surface of the pillar.

17. The integrated circuit ofclaim 15, wherein the bit line is disposed between the body of the access transistor and the substrate.

18. The integrated circuit ofclaim 15, wherein the first source/drain region of the access transistor includes a less heavily doped portion and a more heavily doped bit line portion.

19. The integrated circuit ofclaim 15, wherein the first source/drain region of the access transistor includes a less heavily doped portion and a more heavily doped bit line portion.

20. The integrated circuit ofclaim 15, wherein the pillar also has opposing third and fourth sides, and the third and fourth sides face a third trench and a fourth trench, respectively.

21. A memory array, comprising:

a substrate;

an array of semiconductor pillars that extend outwardly from a working surface of the substrate, each pillar having a plurality of sides including a first side opposite a second side, each pillar having a first source/drain region, a body formed vertically on the first source/drain region, and a second source/drain region formed vertically on the body, the first and second source/drain regions and the body of each pillar forming portions of an access transistor for a memory cell, the array including at least one column including a first pillar, a second pillar separated from the first pillar by a first trench, and a third pillar separated from the second pillar by a second trench, the first and second sides of the second pillar facing the first and second trenches, respectively;

a bit line electrically interconnecting one of the first and the second source/drain regions of the first pillar, the second pillar and the third pillar;

a pair of electrically-isolated first word lines on opposite sides of the first trench, a first word line of the pair of first word lines disposed adjacent to the second side of the first pillar to form a gate for the first pillar, and a second word line of the pair of first word lines disposed adjacent to the first side of the second pillar to form a first gate for the second pillar; and

a pair of electrically-isolated second word lines on opposite sides of the second trench, a first word line of the pair of second word lines disposed adjacent to the second side of the second pillar to form a second gate for the second pillar, and a second word line of the pair of second word lines disposed adjacent to the first side of the third pillar to form a gate for the third pillar.

22. The memory array ofclaim 21, wherein the bit line is disposed between the bodies of the first pillar, the second pillar and the third pillar, and the substrate.

23. The memory array ofclaim 21, wherein the first source/drain region of each of the first pillar, the second pillar and the third pillar includes a more heavily doped bit line portion.

24. The memory array ofclaim 21, further comprising a storage capacitor coupled to one of the first and second source/drain regions of each of the first, second and third pillars.

25. A memory device, comprising:

a substrate;

an array of memory cells arranged in columns and rows on the substrate, adjacent pairs of columns separated by column trenches and adjacent pairs of rows separated by row trenches, each cell comprising a storage capacitor and an access transistor, each access transistor including a semiconductor pillar forming a vertically-stacked body region and first and second source/drain regions, and including first and second gates disposed adjacent to opposing sides of the pillar;

a plurality of bit lines, each bit line electrically interconnecting one of the first and second source/drain regions of each of the pillars in one of the columns of the array; and

a pair of electrically-isolated word lines disposed in each of the row trenches, the first gate of each access transistor controlled by one word line of the pair of word lines disposed in the row trench on one side of the pillar, the second gate of each access transistor controlled by one word line of the pair of word lines disposed in the row trench on the opposing side of the pillar.

26. The memory device ofclaim 25, wherein each bit line is disposed between the body regions of the pillars in the respective column of the array and the substrate.

27. A memory array, comprising:

a substrate;

a plurality of semiconductor pillars that extend outwardly from a working surface of the substrate, each pillar having a plurality of sides including a first side opposite a second side, each pillar having a first source/drain region, a body formed vertically on the first source/drain region, and a second source/drain region formed vertically on the body, the first and second source/drain regions and the body of each pillar forming portions of an access transistor for a memory cell, the pillars being arranged in an array including a plurality of columns separated by first trenches and a plurality of rows separated by second trenches extending orthogonally to the first trenches, the first and second sides of the pillars in each row of pillars facing adjacent second trenches;

a plurality of bit lines, each bit line electrically interconnecting one of the first and the second source/drain regions of each of the pillars in one of the columns of the array, each bit line being at least partially disposed in the substrate; and

a pair of electrically-isolated word lines on opposite sides of each second trench, a first word line of each pair disposed adjacent to the first sides of the pillars in the row of pillars on a first side of the respective second trench to form a first gate for those pillars, a second word line of each pair disposed adjacent to the second sides of the pillars in the row of pillars on a second side of the respective second trench to form a second gate for those pillars, whereby each pillar has first and second gates associated with the first and second sides of that pillar, respectively.

28. A memory device, comprising:

a substrate;

an array of memory cells arranged in columns and rows on the substrate, adjacent pairs of columns separated by column trenches and adjacent pairs of rows separated by row trenches, each cell comprising a storage capacitor and an access transistor, each access transistor including a semiconductor pillar forming a vertically-stacked body region and first and second source/drain regions, and including first and second gates disposed adjacent to opposing sides of the pillar;

a plurality of bit lines, each bit line electrically interconnecting one of the first and second source/drain regions of each of the pillars in one of the columns of the array, each bit line being at least partially disposed in the substrate; and

a pair of electrically-isolated word lines disposed in each of the row trenches, the first gate of each access transistor controlled by one word line of the pair of word lines disposed in the row trench on one side of the pillar, the second gate of each access transistor controlled by one word line of the pair of word lines disposed in the row trench on the opposing side of the pillar.

Images